search

Josh-Lee1 / eBird-Fire-Index

BEES3041 Big Data Project
0 stars 0 forks source link

Range Sizes #20

Closed Josh-Lee1 closed 3 years ago

Josh-Lee1 commented 3 years ago

Hey @coreytcallaghan I'm pretty sure the taxonomies match up ok but missing 5 species (Golden Whistler, Silvereye, Yellow-tailed Black-cockatoo, Shining Bronze-cuckoo and Rose Robin).

Is it possible that these species are missing from the data or are you sure they are in there?

Thanks, Josh

coreytcallaghan commented 3 years ago

Hmm...

Not sure. They should be there. According to here, it appears they have range maps: http://datazone.birdlife.org/species/factsheet/rose-robin-petroica-rosea

Maybe go through the 5 and see if the have a different name or something and that's why they are missing?

Josh-Lee1 commented 3 years ago

From googling around it doesn't seem like these species have any other common names. Is there a version of this dataset with scientific names?

coreytcallaghan commented 3 years ago

Ah, I see. Um probably somewhere.

But if it is only five species, the easiest thing is to probably just add them and manually grab the data for those five species from the datazone: http://datazone.birdlife.org/species/search

Like here:

image

Josh-Lee1 commented 3 years ago

Easy. Thanks.

Josh-Lee1 commented 3 years ago

This data doesn't seem to match the file that you originally passed on to me. I've checked a few species and numbers look different after converting to miles to km. I'm looking at: Extent of Occurrence breeding/resident (km2)

Josh-Lee1 commented 3 years ago

Yeah I think the numbers are off. If I fill in the five with Extent of Occurrence breeding/resident (km2) the lm is non significant and three of the five are major outliers. I ran it without the five to compare and P = 0.0006. So looks like it will be a good result but need to find the right numbers.

coreytcallaghan commented 3 years ago

Okay. Let me try to have a look at it today or tomorrow.

On Wed, Mar 24, 2021, 1:42 AM Josh-Lee1 @.***> wrote:

Yeah I think the numbers are off. If I fill in the five with Extent of Occurrence breeding/resident (km2) the lm is non significant and three of the five are major outliers. I ran it without the five to compare and P = 0.0006. So looks like it will be a good result but need to find the right numbers.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Josh-Lee1/eBird-Fire-Index/issues/20#issuecomment-805382434, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGWSEJVDCOLFNW4XVEQEVLDTFEYLZANCNFSM4ZUKD4NQ .

coreytcallaghan commented 3 years ago

I just sent you a different range size dataset and Rose Robin is in there (by its scientific name), and also doesn't match the online eoo so they must be different.

Go with what is in the most recent data, but hopefully they are close to the other dataset I sent through.

Let me know how you go.

Josh-Lee1 commented 3 years ago

RS

image

Josh-Lee1 commented 3 years ago

Yeah that data seemed to have everyone in it after I fixed a few name discrepancies. Do you think this data looks like it needs transforming?

coreytcallaghan commented 3 years ago

Sweet. Probably log10 for range size, yeah.

On Wed, Mar 24, 2021, 6:06 AM Josh-Lee1 @.***> wrote:

Yeah that data seemed to have everyone in it after I fixed a few name discrepancies. Do you think this data looks like it needs transforming?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Josh-Lee1/eBird-Fire-Index/issues/20#issuecomment-805500246, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGWSEJXKD72K4RDEDPHE4DLTFFXJRANCNFSM4ZUKD4NQ .

Josh-Lee1 commented 3 years ago

RSlog10

This looks wrong... Is ggplot (blah blah)+ scale_x_log10() what I'm supposed to do?

Also aren't we supposed to transform before the model? if so then we didnt log transform bird size...

Josh-Lee1 commented 3 years ago

image

Cause log transforming makes size significant... haha oops. How big of an issue is this for us?!?

coreytcallaghan commented 3 years ago

Hmm. Not a big issue. See what WKC thinks! Range size looks weird, so maybe best not to log transform. Need to look at the model fit etc.

And yeah, I usually do +scale_x_log10() as well.

Good work though, and either way, it will be good to include this result and figure as an additional panel on the previous one I think.

Let's set up a time to chat through all these things perhaps and then you can go forth with the revisions! So close!

On Wed, Mar 24, 2021 at 6:49 AM Josh-Lee1 @.***> wrote:

[image: image] https://user-images.githubusercontent.com/59299694/112261675-6de0f180-8cc0-11eb-9728-c5d5c9782658.png

Cause log transforming makes size significant... haha oops. How big of an issue is this for us?!?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Josh-Lee1/eBird-Fire-Index/issues/20#issuecomment-805519731, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGWSEJVPUK77BML5ZB2DWJ3TFF4NNANCNFSM4ZUKD4NQ .